A conserved regulatory mechanism mediates the convergent evolution of plant shoot lateral organs

Land plant shoot structures evolved a diversity of lateral organs as morphological adaptations to the terrestrial environment, with lateral organs arising independently in different lineages. Vascular plants and bryophytes (basally diverging land plants) develop lateral organs from meristems of sporophytes and gametophytes, respectively. Understanding the mechanisms of lateral organ development among divergent plant lineages is crucial for understanding the evolutionary process of morphological diversification of land plants. However, our current knowledge of lateral organ differentiation mechanisms comes almost entirely from studies of seed plants, and thus, it remains unclear how these lateral structures evolved and whether common regulatory mechanisms control the development of analogous lateral organs. Here, we performed a mutant screen in the liverwort Marchantia polymorpha, a bryophyte, which produces gametophyte axes with nonphotosynthetic scalelike lateral organs. We found that an Arabidopsis LIGHT-DEPENDENT SHORT HYPOCOTYLS 1 and Oryza G1 (ALOG) family protein, named M. polymorpha LATERAL ORGAN SUPRESSOR 1 (MpLOS1), regulates meristem maintenance and lateral organ development in Marchantia. A mutation in MpLOS1, preferentially expressed in lateral organs, induces lateral organs with misspecified identity and increased cell number and, furthermore, causes defects in apical meristem maintenance. Remarkably, MpLOS1 expression rescued the elongated spikelet phenotype of a MpLOS1 homolog in rice. This suggests that ALOG genes regulate the development of lateral organs in both gametophyte and sporophyte shoots by repressing cell divisions. We propose that the recruitment of ALOG-mediated growth repression was in part responsible for the convergent evolution of independently evolved lateral organs among highly divergent plant lineages, contributing to the morphological diversification of land plants.


Introduction
During 470 million years of evolution, the body plans of land plants diversified independently among the gametophyte and sporophyte life stages of different plant groups. In extant bryophytes, early-diverging land plants, the gametophyte is the dominant phase of the life cycle [1,2]. The gametophyte comprises an apical-basal axis with an apical stem cell and forms structures in which gametes develop (antheridiophores and archegoniophores). In contrast, the sporophyte is dominant in extant vascular plants. The sporophyte comprises an axial system (shoots or stems growing along apical-basal axes) that develops from an apical meristem and forms structures in which haploid spores develop. Therefore, in different plant lineages, gametophytes and sporophytes develop axial systems that are produced by apical meristems [3][4][5].
Extant bryophytes and vascular plants develop lateral organs on gametophytes and sporophytes, respectively. Apical meristems maintain stem cell activity at their center and iteratively generate lateral organs at the meristem periphery. The spatiotemporal differences in cell division and expansion in lateral organs contribute to the morphological diversity of shoot structures in land plants [6][7][8][9][10][11].
The liverwort Marchantia polymorpha is a bryophyte that forms a gametophyte axis that undergoes indeterminate planar growth in the form of a flattened mat of tissue, called a thallus. The thallus exhibits strong dorsoventrality; gemma cups, gemmae, and air chambers develop on the dorsal side, whereas rhizoids and ventral scales are formed on the ventral side ( Fig 1A) [12][13][14][15][16][17]. Ventral scales cover bundles of rhizoids that run along the underside of the thallus and facilitate water and nutrient transport over the ventral surface of the thallus ( Fig  1B) [17]. In the leafy liverworts, photosynthetic leaves arise next to a tetrahedral single stem cell (apical cell). By contrast, M. polymorpha does not develop photosynthetic leaves. Instead, the ventral scales alternately develop on the left and right sides of the wedge-shaped apical cell on the ventral surface in the apical notch near the growing tip of the thallus. The flattened form, single-cell thickness, and bilateral symmetry of the Marchantia scales resemble the leaves in leafy liverworts ( Fig 1C) [16,17]. The ventral scales of M. polymorpha are hypothesized to be homologous to the photosynthetic leaves of the basally diverging leafy liverworts [18,19].
The fossil record indicates that the shoot of the earliest known land plants comprised branching stems without lateral, determinate organs [20,21]. Subsequently, determinate lateral organs, which develop from the sides of apical meristems, evolved. The earliest example of such a lateral organ is the microphyll that developed on the stems of the sporophyte of Baragwanathia longifolia, a lycophyte, which first appears in the fossil record in the late Silurian [22,23]. No lateral organs are known from the gametophytes of early bryophytes from the Silurian or Devonian; these likely arose subsequently and are found in extant bryophytes. The acquisition and modification of different lateral organ types are likely to have been morphological adaptations to the terrestrial environment that increased photosynthetic efficiency, gas exchange, and water transport [7,[24][25][26].
Mechanisms controlling lateral organ development are well described in angiosperms such as rice and Arabidopsis. However, little is known about the mechanisms that regulate lateral organ development among bryophytes. Therefore, we carried out a forward genetic screen for mutants with defective lateral organ development in the liverwort M. polymorpha to define mechanisms that control lateral organ development in this species. Comparing the roles of the genes that control lateral organ development in liverworts and angiosperms allows the identification of mechanisms that were involved in the independent evolution of analogous lateral organs during land plant evolution [7,[24][25][26].

M. polymorpha LATERAL ORGAN SUPPRESSOR 1 specifies lateral organ identity during vegetative growth
We isolated two mutants, vj99 and vj86, that produced abnormal green outgrowths from a population of 105,000 transfer DNA (T-DNA)-transformed M. polymorpha (Fig 1D-1G; S1A and S1B Fig). vj99 and vj86 thalli were hyponastic, bending upward at the thallus margins, unlike wild type (WT) (Fig 1D and 1E; S1C and S1D Fig). A single T-DNA was inserted into the gene Mapoly0028s0118 in vj99 and vj86, suggesting that defective function of Mapoly0028s0118 was responsible for the green outgrowths (S1E Fig). To test this hypothesis, we generated independent mutations in the Mapoly0028s0118 gene by homologous recombination. Mutants of Mapoly0028s0118 generated by targeted deletion developed similar phenotypes to those of the vj99 and vj86 mutants (S1F and S1G Fig). To verify that a defect in Mapoly0028s0118 was responsible for the green outgrowth, we transformed mutant vj99 with a genomic fragment that includes the full-length Mapoly0028s0118 gene. Transformation of the Mapoly0028s0118 genomic fragment into vj99 mutants restored WT development, demonstrating that loss of Mapoly0028s0118 function causes the vj99 phenotype (S1H-S1K Fig). Phylogenetic analysis indicated that Mapoly0028s0118 belongs to the Arabidopsis LIGHT-DEPENDENT SHORT HYPOCOTYLS 1 (LSH1) and Oryza G1 (ALOG) protein family (S1L Fig). The proteins in this family contain a DNA-binding domain with weak transcriptional activity [27][28][29]. We named this gene M. polymorpha LATERAL ORGAN SUPPRESSOR 1 (MpLOS1). In addition to the abnormal green outgrowths, gemma cup spacing was abnormal in the Mplos1-1 (vj99) mutant; the distance between neighboring gemma cups was much shorter than in the WT (Fig 1F and 1G).  To more precisely define the nature of the green outgrowths, we performed a phenotypic analysis of the Mplos1-1 mutant. Outgrowths emerged from the ventral side of the thallus near the thallus margins and extended beyond the thallus margin in the Mplos1-1 mutant (Fig 2A-2D). These outgrowths resembled ventral scales in a number of ways. They developed in pairs near the apical notch (Fig 2E and 2F; S2A and S2B Fig). They were in general a single cell layer thick, although the outgrowths located near the apical notch occasionally consist of several cell layers (Fig 2G-2I). Outgrowths located near the apical notch also tended to pile up on one another at the edge of the ventral surface (Fig 2G and 2I). Furthermore, although outgrowths developed, no ventral scales formed on Mplos1-1 mutants ( Fig  2C and 2D), suggesting that the outgrowths are modified ventral scales. Taken together, these data suggested that the outgrowths formed from the ventral thallus on Mplos1-1 mutants were related to ventral scales. Although similar to ventral scales, these outgrowths differ in a number of characteristics. The abnormal outgrowths formed in Mplos1-1 mutants were greener than typical ventral scales (Fig 2J and 2K). There were more cells in outgrowths than in WT scales (Fig 2G-2I; S2C-S2F Fig). Moreover, the mutant chloroplasts were larger than in WT, and there were more thylakoid membranes in the mutants than in WT (Fig 2L and 2M; S2I and S2J Fig). Rhizoids never differentiated in the green outgrowths of the Mplos1-1 mutants, unlike those in WT ventral scales (S2G and S2H Fig). Taken together, these observations indicated that MpLOS1 plays crucial roles in specifying lateral organs as ventral scales, in which MpLOS1 inhibits cell division and chloroplast differentiation.

MpLOS1 activity is required for the maintenance of meristem activity
The WT thallus comprises an apical-basal axis produced by the activity of apical stem cells. The thallus undergoes periodic bifurcation, and gemma cups develop along the midline of the dorsal surface. When the WT thallus bifurcates, a notch containing an apical stem cell forms on each of the two new apical-basal axes ( Fig 3A). This process, the duplication of apical notches and the subsequent growth of thalli, is termed "branching." Upon branching, adjacent apical notches are initially pushed away by the growth of tongue-like tissues, called central lobes, and subsequently further separated concomitant with the growth of the thallus (Fig 3A) [30]. Gemma cups initiate from dorsal merophytes, clones derived from the cell that are cut off from the dorsal surface of the apical cell [17], and they are regularly spaced along the dorsal midline of each thallus.
Gemma cups were more densely arranged along the dorsal surface of Mplos1-1 mutants than in WT ( Fig 1G). This suggested that the mutants had defects in gemma cup differentiation or axis development, or both. To address whether MpLOS1 is involved in bifurcation or gemma cup differentiation, we analyzed the number of apices and gemma cups in Mplos1-1 mutants during cultivation. To count meristems, we imaged expression of the promoter M. polymorpha YUCCA2: ß-glucuronidase (proMpYUC2:GUS) construct, which is preferentially expressed in notches [31]. The number of apical notches expressing GUS was not significantly different between WT and Mplos1-1 mutants until day 7 of cultivation, although subsequently, fewer GUS-expressing apical notches were detected in the Mplos1-1 mutants compared with WT ( Fig 3B-3D). This indicates that bifurcation occurs normally, at least in the early stages of development. In contrast, the density of apical notches in Mplos1-1 mutants was higher than WT at 3 weeks (S3A and S3B Fig). Importantly, there is no clear difference in the number of gemma cups between WT and Mplos1-1 mutants at day 17 of cultivation, although subsequently, fewer gemma cups with higher density were formed in Mplos1-1 mutants (S3B and S3C Fig). These data suggested that the onset of bifurcation, as well as gemma cup differentiation, is not affected and that the separation process of each apical notch is compromised in Mplos1-1 mutants. The lower number of GUS-positive notches and gemma cups in Mplos1-1 mutants after prolonged cultivation may be due to a secondary effect of slow thallus growth or technical limitations of counting densely clustered apices and small-sized immature gemma cups.
The separation of apical notches is dependent on the division and expansion of cells between notches, where central lobes develop (Fig 3A). We reasoned that defective cell division in apical notches and/or central lobes in Mplos1-1 mutant thalli would lead to defects in apical notch separation. We analyzed the cell division activity of Mplos1-1 mutant gemmalings during 7 days' cultivation by applying a 3-hour pulse of 5-ethynyl-2 0 -deoxyuridine (EdU), a thymidine analog that is incorporated into cells during DNA replication (Fig 3E-3G) [32]. In early development (days 1-3), the number of cells labeled by EdU was indistinguishable between the WT and Mplos1-1 mutants ( Fig 3G). However, beginning from day 3, the incorporation of EdU was lower in Mplos1-1 mutants than in WT, which suggested that cell division is reduced in Mplos1-1 mutants compared with WT ( Fig 3E-3G). To determine whether the cell divisions that led to the separation of the apical notches during branching occur in the apical notches themselves or within the central lobes after they emerge, we measured EdU incorporation in these areas during WT development. We found that many cells incorporated EdU in the incipient central lobe as it forms in the apical notch region (S3D Fig). In contrast, very little cell division is found within the central lobe once it has emerged (S3E Fig). This indicates that cells in central lobes are mainly supplied from the apical notches, whereas the contribution of cell division activity within the central lobe for separating duplicated apical notches is low. Altogether, these data suggest that the rate of cell division in the apical notch is lower in the mutant than in WT, leading to an impaired separation of apical notches during branching.
We also compared the cellular organization of apical meristems of Mplos1-1 mutants and WT. The WT apical meristem comprised a single wedge-shaped apical cell and surrounding merophytes (a group of clonally related cells resulting from sequential cell divisions in a single derivative of the apical cell of a meristem), in which the lateral merophytes and the apical cell display identical shapes ( Fig 3H) [17]. In contrast, there were many wedge-shaped cells in Mplos1-1 mutants, in contrast to the single apical cell of WT ( Whereas the expression of proMpYUC2:GUS was restricted to a small area of the WT apical notch, staining was more dispersed in Mplos1-1 mutants (Fig 3J and 3K). Occasionally (3 out of 20 gemmalings at 14 days' cultivation), apical meristems were aborted in Mplos1-1 mutants (S4I Fig, dotted boxes), a phenomenon not observed in WT in our conditions. These data suggest that MpLOS1 is required for the maintenance of apical meristems. These data also support the hypothesis that Mplos1-1 mutants fail to separate apical notches because of defects in cell proliferation in the apical notches. Gemma cup differentiation as well as bifurcation initiate as in WT, but then subsequent defective meristem activity causes defective axis expansion, resulting in the development of a higher density of gemma cups and apical notches in the Mplos1-1 mutant thallus.

MpLOS1 is expressed in lateral organs but not in apical cells
To define the spatial expression patterns of MpLOS1, we established a line that expressed GUS under the control of 5 0 and 3 0 regulatory elements that were used in the complementation analysis of Mplos1-1 mutants (S1E Fig). In 4-day-old gemmalings, GUS staining was detected in notches and rhizoids (Fig 4A). Weak signal was observed elsewhere in growing thalli (Fig 4B  and 4C). The developing ventral scales in the ventral region of the apical notch stained the strongest (Fig 4B-4F). Staining extended over the entire young ventral scale and the basal region of old ventral scales (Fig 4D and 4E). No signal was detected in the oldest ventral scales (Fig 4D and 4E). The expression of MpLOS1 in ventral scales is consistent with the phenotypic defects seen in these organs in the mutant, further strengthening our hypothesis that MpLOS1 is required for the normal development of ventral scales. We also expressed functional proM-pLOS1:enhanced green fluorescent protein (eGFP)-MpLOS1 constructs in Mplos1-1 mutants (S1H, S1I and S1K  apical cells or lateral merophytes despite the defect in apical meristem morphology and maintenance in Mplos1-1 mutants (Fig 4G-4I). These findings suggest that MpLOS1 mediates the maintenance of apical meristems non-cell autonomously, although we cannot exclude the possibility that MpLOS1 proteins below the level of detection in the apical meristems maintain meristem activity (Fig 4I).

MpLOS1 specifies lateral organ identity during reproductive growth
M. polymorpha produces an umbrellalike gametangiophore (antheridiophore or archegoniophore) that bears antheridia or archegonia during reproductive growth [17]. The gametangiophore is a vertically growing thallus branch [17], and we reasoned that gametangiophore development might be defective in Mplos1-1 mutants. The antheridial receptacles of male Mplos1-1 plants were smaller than WT, and unlike in the WT, antheridia were frequently exposed (S6A- S6C Fig) The archegonial receptacle of female M. polymorpha is highly lobed, with finger-like structures called digitate rays (Fig 5A). The archegonial receptacle lacks the rows of typical ventral scales that develop in antheridiophores. Instead, a pair of specialized scalelike structures called involucres, which are larger than ventral scales, develop between each digitate ray and enclose the archegonia cluster (Fig 5C and 5D) [17]. In female Mplos1-2 mutants (vj86), large leaf-like structures developed as in antheridiophores (Fig 5E). Moreover, more than two involucre-like structures differentiated between each digitate ray (Fig 5C-5F). Importantly, these involucrelike structures resembled ventral scales in their arrangement in several rows (Figs 1B, 1C, 5G and 5H). This suggests that loss of MpLOS1 function results in the transformation of involucres into more scalelike structures. GUS staining was also detected in immature involucres but not in mature involucres in proMpLOS1::GUS archegoniophores (Fig 5I; S6K Fig), accompanied by staining of all parts of the archegonia including eggs, collars, and venters (Fig 5I; S6K and S6L Fig) [17]. These data suggest that ventral scales are transformed into involucres in a MpLOS1-dependent manner upon the transition from vegetative to reproductive growth, in which MpLOS1 inhibits the growth of two rows of ventral scales, resulting in the formation of a single pair of involucres.

Protein function of ALOG proteins are conserved between Marchantia and rice
Mutation of long sterile lemma (G1), a member of the ALOG gene family in rice, results in the enlargement of sterile lemmas, a small leafy lateral organ in the rice spikelet (flower clusters in grass species, basic unit of inflorescence, consisting of one or more flowers), which is interpreted as a homeotic transformation of a sterile lemma into a lemma [29]. Similarly, Mplos1-2 mutants displayed defects in lateral organ specification that we interpret as the transformation of involucres into ventral scalelike structures. To determine whether the M. polymorpha protein could rescue the homeotic transformation of the rice mutant, we expressed MpLOS1 in rice g1 mutants. Expression of MpLOS1 restored the WT short sterile lemma phenotype (Fig 6A-6C). This suggests that the protein functions of the ALOG proteins have been conserved since the time that M. polymorpha and rice last shared a common ancestor, which likely lacked lateral organs. It further suggests that ALOG family proteins were independently co-opted to specify sporophytic function in the lineage giving rise to rice and gametophytic functions in the lineage giving rise to liverworts when each originated the evolutionary novelty of lateral organs.

Land plant ALOG proteins regulate lateral organ development and meristem activity
Here, we report the discovery that MpLOS1, a member of the ALOG protein family, controls both lateral organ development and apical meristem activity in M. polymorpha. MpLOS1 represses the growth of different lateral organs, including ventral scales and involucres, and MpLOS1 expression was detected early in the development of these lateral structures. These data indicate that the gene is required for normal lateral organ development. Furthermore, MpLOS1 activity is required for apical meristem maintenance. However, MpLOS1 is not expressed in the apical meristems or surrounding cells. We propose that MpLOS1 cell autonomously regulates lateral organ development but non-cell autonomously regulates apical meristem maintenance (Fig 4I).
The role of ALOG proteins in meristem maintenance is conserved between monocots and dicots. Oryza sativa TAWAWA1(OsTAWAWA1) and Solanum lycopersicum TERMINAT-ING FLOWER (SlTMF) (the tomato TAW1 homolog) proteins, members of the ALOG gene family in rice and tomato, respectively, repress maturation of meristems during reproductive growth [28,33,34]. Although the angiosperm genes control meristem development, neither SlTMF, Arabidopsis thaliana LSH3 (AtLSH3) (the Arabidopsis TAW1 homolog), nor OsTAW1 proteins are expressed in apical meristems. Instead, they are expressed at lateral organ boundaries [27,33,35]. Taken together, these data from a diversity of land plants suggest that although the ALOG genes act cell autonomously during the development of lateral organs, they act non-cell autonomously to control meristem development. It remains unclear how this might operate, but there is evidence from angiosperms that lateral organ development is required for meristem maintenance [9,36,37].
Taken together with our discovery that MpLOS1 is required non-cell autonomously for meristem maintenance in M. polymorpha, this means that the evolutionarily conserved ALOG family proteins control apical meristems in divergent plant lineages, in which the apical meristems are found in different phases of the life cycle. We propose that this mechanism for controlling shoot meristematic activity was already present in the last common ancestor of Marchantia and rice.

Conserved ALOG proteins negatively regulate lateral organ growth
We discovered that MpLOS1 specifies lateral organ identity by negatively regulating the lateral organ outgrowth; involucres are transformed into ventral scalelike structures during reproductive growth in Mplos1-2 mutants (Fig 5G and 5H). The rice homolog, G1, also represses the development of lateral organs to specify the sterile lemmas. Loss-of-function mutations in OsG1 result in the transformation of small sterile lemmas into large lemmas ( Fig  6A and 6B) [29]. Similarly, in tomato, Sltmf mutants display a similar transformation, in which sepals develop leaf characteristics [33]. The conserved function of rice, tomato, and M. polymorpha LOS1 homologs suggests that the role of ALOG proteins in repression of lateral organ development is ancient. This functional conservation among divergent taxa of land plants suggests two alternative hypotheses regarding the evolution of lateral organs.
According to the first hypothesis, the common ancestor of liverworts and the seed plants developed lateral organs whose growth was controlled by one or more ALOG genes. Because lateral organs are present in the gametophyte and the sporophyte phase in liverworts and seed plants, respectively, under this scenario, the common ancestor may have had lateral organs in both the gametophyte and the sporophyte phase. These structures would then have subsequently diverged morphologically during the course of land plant evolution. However, this hypothesis is not well supported by the fossil record. The earliest polysporangiophyte and tracheophyte fossils we know of, such as Aglaophyton and Cooksonia, possessed simple axes devoid of lateral organs [20,21]. This suggests that their last common ancestor with the liverworts likewise lacked lateral organs.
An alternative hypothesis, which is in better accord with the fossil record, is therefore that the common ancestor of liverworts and the seed plants lacked lateral organs in the gametophyte and sporophyte phase but possessed an ALOG-mediated mechanism for controlling some other developmental process. The ALOG-dependent growth-repression mechanism was subsequently recruited independently during the evolution of lateral organs in the separate lineages leading to the liverworts and seed plants. If the last common ancestor did not develop lateral organs, and because ALOG function regulates lateral organ development in both liverworts and seed plants, we suggest that ALOG function was recruited independently during the evolution of lateral organs in different lineages of land plants. The original ALOG function prior to these independent recruitment events may have been to control some aspect of apical meristem activities because growth by apical meristems is a shared characteristic of land plants [38] and this ALOG function is found in both liverworts and seed plants. The recruitment of ALOG function during the independent evolution of lateral organs thus provides a molecular mechanism for the convergent evolution of growth repression in lateral organs.

ALOG proteins may mediate diversification of lateral organs during plant evolution
It has been suggested that the repressive activity of the OsG1 gene on growth led to the evolution of the rice spikelet [29]. Loss-of-function Osg1 mutations revert the sterile lemma into a larger leafy structure, which has been interpreted as similar to a hypothetical ancestral structure (Fig 6A and 6B) [29]. According to this model, the formation of a pair of lower lemmas subtending the floret (rice flower surrounded by two bracts; the external lemma and internal palea) was the ancestral state. Then, during the evolution of rice, OsG1 activity was co-opted to repress the development of the lower lemma, which is now much reduced in size in modern rice compared with the ancestral state, resulting in the formation of the sterile lemma. It is formally possible that MpLOS1 may also have played a similar role in the evolution of lateral organs in liverworts. Several liverwort taxa with thalloid form are suggested to have evolved independently from ancestral leafy liverworts, in which leaves are hypothesized to be transformed into nonphotosynthetic ventral scales with reduced growth during this evolutionary transition [18,19]. We found that MpLOS1 is involved in the specification of lateral organ identities by inhibiting cell division and chloroplast differentiation and that the loss of its function leads to the formation of chlorophyll-containing photosynthetic tissues (Fig 2J and 2K). These green appendages are in fact similar to the green-colored photosynthetic scales formed in the Treubiaceae family of liverworts, whose semithalloid form has been interpreted as an evolutionary transition state between the leafy and thalloid form [39][40][41]. Therefore, MpLOS1 function may also be associated with the evolution of the thallose body form by repressing leaf growth in ancestral leafy liverworts in the same way that OsG1 suppresses lower lemma development during rice spikelet evolution. It is possible that morphological modification of lateral organs is controlled by the spatial and temporal differences in expression levels of ALOG family genes, and this would provide a mechanism for the establishment of morphological diversification in lateral organs that develop on shoots during land plant evolution.

Conclusion
We demonstrate that MpLOS1, a member of the ALOG gene family, plays a role in integrating meristem activity and lateral organ differentiation in M. polymorpha. MpLOS1 acts by repressing lateral organ growth and is required for meristem maintenance. Because ALOG proteins from angiosperms also repress lateral organ growth and are required for meristem maintenance, and these functions were rescued by MpLOS1, we conclude that protein functions of ALOG family proteins are conserved among these taxa and acted in their last common ancestor. We hypothesize that ALOG genes were co-opted to execute the morphological modification of analogous lateral organs during land plant evolution and contributed to the diversification of lateral organs in shoot systems during the course of land plant evolution.

Phylogenetic analysis
Phylogenetic analysis was performed as described by Bowman and colleagues [3]. Protein sequences were collected using the Marchantia genome portal site MarpolBase (http:// marchantia.info). Multiple sequence alignments were performed using the MUSCLE program [46] contained in the Geneious software (https://www.geneious.com). Gaps were removed by using Strip Alignment Columns in the Geneious package, and phylogenetic analyses were performed using PhyML (http://www.atgc-montpellier.fr/phyml/).

Plasmid construction
For constructing the proMpLOS1:MpLOS1 plasmid that complements the Mplos1 mutants, a MpLOS1 genomic fragment with a 10-kb upstream region and a 3-kb downstream region was amplified by PCR using Prime STAR GXL polymerase (TaKaRa, Shiga, Japan) and subcloned into pENTR/D-TOPO (Thermo Fisher Scientific, Waltham, Massachusetts, United States), which was subsequently integrated into pMpGWB101 by a Gateway LR reaction [47]. pENTR/ D-TOPO that included the proMpLOS1:MpLOS1 complement fragment was modified to establish proMpLOS1:eGFP-MpLOS1 and proMpLOS1:GUS plasmids. A PCR-amplified eGFP coding sequence was inserted in frame with the 5 0 end of the MpLOS1 coding sequence by the In-Fusion cloning reaction (TaKaRa, Shiga, Japan) to generate the proMpLOS1:eGFP-MpLOS1 plasmid. The coding sequence of MpLOS1 in the proMpLOS1:MpLOS1 complement fragment was replaced with a PCR-amplified GUS coding sequence by the In-Fusion cloning reaction to generate proMpLOS1:GUS plasmids. pENTR/D-TOPO vectors that included proMpLOS1:eGFP-MpLOS1 or proMpLOS1:GUS fragments were subsequently integrated into pMpGWB101 by the Gateway LR reaction. For overexpression of MpLOS1 in rice g1 mutants, a plasmid carrying pro35S::MpLOS1 was constructed. The MpLOS1 genomic coding sequence was amplified by PCR and subcloned into pENTR/D-TOPO, which was subsequently integrated into pGWB2 by a Gateway LR reaction [48].

Histochemical GUS staining
GUS staining was performed as described by Naramoto and colleagues [49], except that 50 mM sodium phosphate buffer was used. Samples were cleared with 70% ethanol and subsequently mounted using a clearing solution (chloral hydrate:glycerol:water, 8:1:2) for direct microscopic observation or dehydrated through a graded ethanol series and embedded in paraffin or Technovit 7100 resin for microtome sectioning.

Plant embedding and sectioning
Plant material was fixed in FAA (45% ethanol:5% formaldehyde:5% acetic acid in water) for embedding in paraffin and Technovit 7100. For paraffin embedding, fixed plant material was dehydrated in a series of ethanol (25%-50%), t-butyl alcohol (10%-75%), and chloroform (20%) solutions and then embedded in Paraplast (McCormick, Baltimore, Maryland, USA). For Technovit 7100 embedding, fixed samples were dehydrated through a graded ethanol series and embedded in Technovit 7100 resin according to the manufacturer's instructions (Heraeus Kulzer, Hanau, Germany). Embedded samples were sectioned on a rotary microtome into a series of vertical transverse and longitudinal sections (thickness of 8 μm for paraffin and 4 μm for Technovit sectioning). The obtained sections were further treated with neutral red dyes as a counterstain for GUS-stained samples or toluidine blue for the other samples. Multi-Mount 480 solution (MATSUNAMI, Osaka, Japan) or Entellan new (MERCK, Darmstadt, Germany) were used as mounting agents to preserve the samples on the slides.

ClearSee treatment and staining of cell walls
Plants were fixed with 4% paraformaldehyde (PFA) in 1× PBS for 1 hour at room temperature under vacuum. Samples were subsequently washed twice with PBS and transferred to ClearSee solution (10% xylitol, 15% sodium deoxycholate, and 25% urea in water) [50]. ClearSee treatment was prolonged until samples became transparent. Cell walls were stained for 1 hour with 0.1% (v/v) calcofluor or with 0.1% (w/v) Direct Red 23 dissolved in ClearSee. Stained samples were washed for at least 30 minutes with ClearSee solution before observation.

EdU uptake experiments
Gemmalings were incubated in 1/2 B5 medium containing 10 μM EdU (Click-iT EdU Alexa Fluor 488 imaging kit; Thermo Fisher, Waltham, Massachusetts, USA) for 3 hours. Samples were fixed with 4% PFA in 1× PBS for 1 hour under vacuum and then washed three times in PBS. Coupling of EdU to the Alexa Fluor substrate was performed according to the manufacturer's instructions. Before observations, samples were cleared with ClearSee solution, and cell walls were subsequently stained by Direct Red 23.

Microscopy
Anatomical features were observed with a light microscope (Olympus BX51) equipped with an Olympus DP71, a light sheet microscope (Zeiss Z.1), or a confocal laser scanning microscope (Olympus FV1000 or Zeiss LSM880). For light microscope observations, a PLAPON 2× objective, a UPlanFl 10× objective, or a UPlanFl 20× objective was used. Light sheet microscope observations were conducted using Lightsheet Z.1 detection optics 5× or Clr Plan-Neofluar 20×. For confocal laser scanning microscopy, cell walls stained by Calcofluor or by Direct Red 23 were excited at 405 nm or 543 nm, respectively, whereas GFP and Alexa 488-labeled EdU were excited at 488 nm. Samples were mounted using ClearSee solution and observed with silicon oil objectives. The 3D reconstruction was done by using Imaris software (BITPLANE, http://www.bitplane.com/). High-resolution images showing ultrastructural details were obtained using an SEM (JEOL JCM-6000Plus NeoScope) and an FESEM (Hitachi SU820).

Data analysis and statistics
Four alleles of the Mplos1 mutants, including Mplos1-1, Mpos1-2, Mplos1-3, and Mplos1-4 displayed identical phenotypes in vegetative growth and, thus, Mplos1-1 mutants were used as a representative allele unless otherwise described. All experiments were repeated at least three times. Representative images were used for the preparation of figures. Statistical analysis was conducted using Excel. For comparing two groups, we used the t test to calculate the p-value. The p-values of the relevant experiments are described in the Supporting information (S2 Data). Materials Design Analysis Reporting checklist is described in the Supporting information (S1 MDAR Checklist).